Intended for healthcare professionals

Feature Readers Respond

What readers thought about a research paper’s approach to compare treatment effects on mortality

BMJ 2018; 362 doi: (Published 17 August 2018) Cite this as: BMJ 2018;362:k3213
  1. José G Merino, US research editor,
  2. Elizabeth Loder, head of research,
  3. Kamran Abbasi, executive editor
  1. The BMJ
  1. jmerino{at}

What is the best way to compare treatment interventions when it is not evident which is standard and which is experimental? A research article on one approach generated a lively discussion of over 20 responses

On 8 February 2016, The BMJ published the research paper “Agreement of treatment effects for mortality from routinely collected data and subsequent randomized trials: meta-epidemiological survey” by Lars Hemkens, Despina G Contopoulos-Ioannidis, and John P A Ioannidis (doi:10.1136/bmj.i493).1 In this study, the authors used a ratio of odds ratios (ROR) approach to compare the treatment effects on mortality seen in observational studies that use routinely collected data (RCD) to that seen in subsequent randomised controlled trials (RCTs). They concluded that RCD studies systematically overestimate the mortality benefits of medical treatments. In the months following the publication, the rapid responses section of was the forum for a lively discussion on the relative merits and weaknesses of the study (to date, over 20 responses have been posted).

Call for retraction

To ensure consistency when studies …

View Full Text

Log in

Log in through your institution


* For online subscription